Reconfigurable intelligent surface (RIS) has gained much traction due to its potential to manipulate the propagation environment via nearly-passive reconfigurable elements. In our previous work, we have analyzed and proposed a beyond diagonal RIS (BD-RIS) model, which is not limited to traditional diagonal phase shift matrices, to unify different RIS modes/architectures. In this paper, we create a new branch of BD-RIS supporting a multi-sector mode. A multi-sector BD-RIS is modeled as multiple antennas connected to a multi-port group-connected reconfigurable impedance network. More specifically, antennas are divided into $L$ ($L \ge 2$) sectors and arranged as a polygon prism with each sector covering $1/L$ space. Different from the recently introduced concept of intelligent omni-surface (or simultaneously transmitting and reflecting RIS), the multi-sector BD-RIS not only achieves a full-space coverage, but also has significant performance gains thanks to the highly directional beam of each sector.We derive the constraint of the multi-sector BD-RIS and the corresponding channel model taking into account the relationship between antenna beamwidth and gain. With the proposed model, we first derive the scaling law of the received signal power for a multi-sector BD-RIS-assisted single-user system. We then propose efficient beamforming design algorithms to maximize the sum-rate of the multi-sector BD-RIS-assisted multiuser system. Simulation results verify the effectiveness of the proposed design and demonstrate the performance enhancement of the proposed multi-sector BD-RIS.

翻译：可重构智能表面（RIS）因其通过近无源可重构元件操控传播环境的潜力而备受关注。在先前工作中，我们分析并提出了超越对角RIS（BD-RIS）模型——该模型不局限于传统对角相移矩阵，旨在统一不同RIS模式/架构。本文开辟了BD-RIS支持多扇区模式的新分支。多扇区BD-RIS被建模为连接至多端口组间可重构阻抗网络的多天线系统。具体而言，天线被划分为$L$（$L \ge 2$）个扇区，并排列成多边形棱柱，每个扇区覆盖$1/L$空间。与近期提出的智能全向表面（或同时发射与反射RIS）不同，多扇区BD-RIS不仅实现全空间覆盖，更因每个扇区的高定向波束而具有显著性能增益。我们推导了多扇区BD-RIS的约束条件及相应的信道模型，其中考虑了天线波束宽度与增益之间的关系。基于所提模型，首先推导了多扇区BD-RIS辅助单用户系统中接收信号功率的标度律，进而提出高效波束赋形设计算法以最大化多扇区BD-RIS辅助多用户系统的总速率。仿真结果验证了所提设计的有效性，并展示了所提多扇区BD-RIS的性能提升。